Greg Landrum

This should definitely have a better error message

> Why not off by default? Firstly because that would be serious change in behavior. Secondly because I believe that it, like the other SMILES and SMARTS extensions the RDKit...

@bp-kelley do you have time to finish this review so that we can get this one in next week's release?

Confirmed. Thanks for the detailed bug report @fioruggiu !

Here's a minimal reproducible of this that does not require tautomer canonicalization: ``` In [2]: m = Chem.MolFromSmiles('CC=C=CC') In [3]: m.GetBondWithIdx(2).SetStereoAtoms(1,4) In [4]: m.GetBondWithIdx(2).SetStereo(Chem.BondStereo.STEREOCIS) In [5]: Chem.MolToSmiles(m) [09:32:30] **** Invariant...

@Mario-Liu : I'm not sure what you're asking about here. The SDF is perfectly valid - so parsing it doesn't generate errors - and the molecule it contains doesn't have...

The RDKit does currently depict undefined double bond stereochemistry if you tell it that the stereochemistry is undefined, so I'm not sure what the intention of this issue is: ...

Thanks for pointing this out. We don't use ReadTheDocs, so that link really should just be removed. PR's are always welcome! :-)

This is related to the way ring closures are handled in SMARTS parsing. That should be changed to make it consistent with the way the SMILES parser does things

Apologies that I didn't end up doing anything here. Thanks for taking it on. I was thinking about the problem incorrectly when I wrote this and no longer believe that...